nace code [13] used here. On the other hand, RBS using MeV ion beams does not have good mass resolution for these chal- cogenide compounds, and RBS ...
Elemental Depth Profiling of Thin Film Chalcogenides Using MeV Ion Beam Analysis 1*
2
2
1
Chris Jeynes , Guillaume Zoppi , Ian Forbes , Melanie J. Bailey , Nianhua Peng
1*
1
2
University of Surrey Ion Beam Centre, Guildford GU2 7XH, UK Northumbria Photovoltaics Applications Centre, Northumbria University, Newcastle upon Tyne, NE1 8ST, UK * Corresponding authors
Abstract: The comprehensive characterisation is one of many technical challenges in the fabrication of photovoltaic devices from novel materials. We show how the application of recent advances in MeV ion beam analysis, providing the selfconsistent treatment of Rutherford backscattering and particle induced X-ray emission spectra, makes a new set of powerful complementary elemental depth profiling techniques available for all thin film technologies, including the chalcopyrite compound semiconductors. We will give and discuss a detailed analysis of a CuInAl metallic precursor film, showing how similar methods are also applicable to other films of interest. I. INTRODUCTION Chalcopyrite-based CuIn1-xGaxSe2 (CIGS) and CuIn1-xAlxSe2 (CIAS) solar cells have achieved the highest level of performance to date for single junction polycrystalline thin film technology [1-3]. Interestingly, the high performance devices were fabricated with materials of a relatively low bandgap (Eg ~ 1.2 eV for 30% Ga or 13% Al substitution respectively). The poor device performance with higher bandgap materials is found to be associated with increased defect density and stronger interfacial recombination when the Ga or Al doping level is increased. These materials are complex, and can be troublesome to fabricate, with many possible fabrication routes. While the most efficient devices so far have been deposited using the co-evaporation method, we have investigated the production of CIAS thin films by a two-stage process: the sputter deposition of Cu/In/Al (CIA) metallic precursor layers followed by annealing in a selenium environment to synthesize the compound [4]. In principle this method promises improved scalability for commercial production compared to other deposition methods, but on the other hand the selenisation technique can
yield unwanted elemental depth profiles due to the binary selenides having different reaction temperatures. So that characterization methods are important for establishing the processes. As a part of our ongoing effort for in-depth analysis of CIA metallic precursors and CIGS and CIAS thin films, we will describe very novel methods of accurate thin film depth profiling using a self-consistent analysis of simultaneously collected spectra from MeV ion backscattering together with the stimulated photon emission from a typical CIA precursor film. II.
DEPTH PROFILING USING ION BEAM ANALYSIS
Conventional thin film depth profiling techniques such as Auger electron or X-ray photo-electron spectroscopy, or SIMS (secondary ion mass spectrometry) are plagued by artefacts including those of interfaces, and SIMS is not quantitative because of the large matrix effects. Other analytical methods such as SEM-EDS (energy dispersive X-ray spectrometry on the scanning electron microscope) have little or no depth resolution and do not work well for these thin films. However, Rutherford backscattering (RBS) is a well-established non-destructive depth profiling technique [5] where the depth resolution comes from the energy loss of the probing beam (such as 4 + 1.5MeV He ) detected after elastic scattering at backward angles from the atomic nuclei of the target; films of CIGS or CIAS of submicron thickness have very convenient energy loss of the primary beam with good depth resolution. Because the RBS elastic scattering cross-section is derived simply from the Coulomb potential [6], and the energy losses of light ion beams in materials are well known [7-9], RBS is an accurate technique suitable for standards work [10-11]. Depth profiles can now be extracted efficiently from RBS spectra (or other related particle scattering spectra) with computer codes validated by an IAEA-sponsored intercomparison exercise [12], including the DataFur-
nace code [13] used here. On the other hand, RBS using MeV ion beams does not have good mass resolution for these chalcogenide compounds, and RBS also has low sensitivity for light elements in a heavy matrix (such as 2 the Al in CuInAl) since the yield goes with Z . Compared to SEM-EDS, particle-induced X-ray emission (PIXE) has orders of magnitude better sensitivity since there is effectively no bremsstrahlung from the primary beam, although it has a similarly poor depth resolution. However, the self-consistent analysis of RBS/PIXE data has recently been introduced, where the resulting analysis has the mass-sensitivity of PIXE combined with the depth-sensitivity of RBS [14-18]. We apply these methods here for the first time to complex thin film PV materials (but see [19]).
a layer structure is calculated through its equivalent excess energy straggling [23-24]. This is not valid for the severe roughness often designed into PV films for maximum light absorption, but IBA spectra from such rough films can also be calculated [25]. NDF has a double scattering calculation [21], and this was included here. The PIXE data were analyzed using the DATTPIXE code of Reis [14] as implemented in NDF by the LibCPIXE module [15]. LibCPIXE interprets characteristic X-ray line areas extracted from the raw data using GUPIX [26-27]. We use a manual procedure in this present work, where we apply cross-section corrections obtained by comparison to X-ray yields calculated (for simplified structures) from GUPIX, using its GUYLS utility (which also gives the fluorescence correction:
